Library / Operations Management Principles Framework
Jump to passage
In this reading

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 11:52:20 UTC · snapshot created 2026-10-03 11:53:41 UTC · last check 2026-10-03 13:50:10 UTC

OPS.20:5.3 - An AI-assisted software operation separates tool use from selection

A software operations group has admitted AI-ReleasePrep-v3 : U.Method and exact AI-ReleasePrep-Description-e4 : U.MethodDescription for one low-risk release-preparation family. A prompt repository, model endpoint, tool rollout, and usage counter are carriers, Systems, or records. They do not show that the on-call population selects or enacts v3.

The group bounds two named on-call rotations and one release interval. It tests one selection predicate: for an eligible low-risk change under the qualified provider and model condition, the responsible human practitioner selects v3 only when its human review, evidence capture, security check, fallback, and release-authority conditions are current. The negative is a logged tool invocation followed by release preparation that omits one required condition.

Credible rivals are that description plus repository access is sufficient, or that a case replay and explicit recognition of the provider/model qualification boundary are needed. Existing eligible-case observations can support a bounded current selection account under the unchanged qualified provider condition. Select a reversible provider-change replay for one rotation only when its attainable result can change the decision enough to warrant the available participants, facilitator, protected window and whole burden, under current release and training authority. Actual handover, release-preparation, review, security and release Work remain separate, as do model calls, generated text, traces, test results, decisions and service consequences.

If the provider or model changes to an unqualified edition, the current branch stops and the named fallback applies. The result can retain one population-specific selection claim while returning the changed-provider branch as unknown or stop. Security, software assurance, model capability, release, employment, and governance consequences remain with their owners; usage count and one successful release establish none of them.